Fourier transform near infrared (FT-NIR) diffuse reflectance spectroscopy as a non-destructive method for the development of calibration models has been investigated as a means of assessing the quality of whole pomegranate fruit. FT-NIR diffuse reflectance spectrometers with different spectral acquisition modes were evaluated using direct contact between the sample and the integrating sphere (IS), or by a contact-less measurement using an optic fibre coupled emission head (EH) to scan fruit over a distance of 170 mm. Fruit weight, firmness and colour components (a*, Chroma, hue angle), total soluble solids (TSS), pH, titratable acidity (TA), sugar to acid ratio (TSS:TA), BrimA, total phenolics, total anthocyanin and vitamin C. The best prediction statistics obtained from calibration models of the EH were firmness (R 2 = 83.0, residual predictive deviation (RPD) = 2.43), fruit colour components (a*: R 2 = 90.9, RPD = 3.34); Chroma: R 2 = 83.0, RPD = 2.43, TSS (R 2 = 78.1, RPD = 2.17), TA (R 2 = 76.8, RPD = 2.12), BrimA (R 2 = 76.2, RPD = 2.08), total phenolics (R 2 = 88.0, RPD = 2.91) and vitamin C (R 2 = 76.2, RPD = 2.06). The best prediction obtained from calibration models of IS were colour component Hue (R 2 = 83.9, RPD = 2.50), TSS:TA (R 2 = 86.8, RPD = 2.72) and total anthocyanin (R 2 = 62.6, RPD = 1.64). Overall, good prediction was observed for both the EH and IS; however, better prediction performance was obtained with the EH which gave the best prediction for 9 of the 13 quality parameters evaluated. These findings have demonstrated that the EH (a contactless option of the Matrix-F) can be implemented as an online tool for the analysis of pomegranate fruit quality. [ABSTRACT FROM AUTHOR]